Vision: a model system of ordered circuit development...

Understanding how complex patterns of connectivity develop in the nervous system is a core Centre activity. The mouse/mammalian visual system provides an excellent model system for this endeavour due to its highly ordered pattern of connections between visual centres. Retinal ganglion cells on the sensory surface project to primary targets within the brain: the visual thalamus and superior colliculus. These primary targets similarly have neurons that project to the visual cortex (a secondary target of visual information), which, in turn, has projections back to the primary targets. Incredibly each of these projections preserve neighbouring visual information by forming ordered (topographic) maps of visual space such that neighbouring areas of visual space project to neighbouring regions within each visual target. Indeed, it is the preservation and coregistering of visual information afford by such maps across several targets that is thought to underlie functional premise for processing visual information.

The advantages of the visual system for the study of circuit development are numerous: anatomically, the topographic connections between visual processing stations are readily visualized and measured; connectivities are amenable to experimental manipulation; the functional readout of the connections are directly and measurably drivable by defined visual paradigms; the functional complexity of neurons increases/changes with different levels of the visual system; and visual function is behaviourally quantifiable.

Work within the lab examines visual circuit development with the aim of: (1) determining the nature and time-course of circuit refinements during critical/defined developmental periods; (2) to dissect the importance of molecular over functional cues associated with orderly circuit formation; and (3) to explore the functional relevance of having such highly ordered circuits.

Measuring map formation in the primary visual circuits

Defining topographic visual maps: "one point in visual space goes to one point in the target and neighbouring points in space go to neighbouring points in the target" In connectivity terms, these becomes more nuanced questions: what extent of retina converges on a single location in the target and at what separation do two target locations receive non-overlapping retinal input?

We are using anatomical tracing techniques to answer these questions, to chart the developmental dynamics of map formation and to examining the mechanisms determining the precision of topographic visual maps in the mouse.

We make small injections of fluorescently labeled latex microspheres into target nuclei and then record the location of retrogradely labeled neurons in the input structure. We have been focusing on the retino-collicular projection. In order to analyse the distribution of labeled ganglion cells, the retina has to be dissected out and flat-mounted.

The flattening and the requirement for relieving cuts introduces topographic distortions. To overcome this colleagues in Edinburgh have developed refolding algorithms to remove cuts and associated distortions. This has allowed more sophisticated analyses to quantify different aspects of map precision and plot the developmental dynamics. On the day of birth, there is no order in the retinal input along the AP axis of the SC but the ML axis shows a degree of order, a level only seen at day nnn for the AP axis. With a thorough description of the emergence of topography, we can begin to dissect the different underlying developmental mechanisms. For instance, genetic perturbation of the normal patterns of spontaneous activity in the retina for the first post-natal week leaves a 'frozen' projection that at P8 resembles that of a P2 normal animal.

Mosaics and maps

The visual system uses both parallel and serial processing strategies to analysis the image. The mammalian retinal contains multiple (more than 20) classes of ganglion cell each of which processes the image in a different way. We are increasingly interested in the how and when these inputs are ordered in the target and, much more challengingly, how they combine to generate new functional characteristics in the post-synaptic neurons.

Several basic questions remain unanswered. Although it's clear that more than 95% of all RGCs in mouse go to the SC (Salinas-Navarro et al., 2009, Vis Res 49:637), the nature of the projection to the thalamus is unclear. A fundamental problem is the mapping of the 2D sheet of ganglion cells map onto the 3D targets. This raises questions about whether ganglion cell classes that display dendritic tiling also display axonal tiling and what factors control the laminar or columnar pattern of terminations of the different classes. Obviously, the ordering of inputs constrains the post-synaptic sampling possibilities and so the functional processing. This is a new direction of research for the lab.

Functional dissection of the visual circuits

The challenge for biomedical research is to devise methods and techniques that can reveal the full complexity of biological systems. In the study of patterned connections, such circuits should be regarded as having an anatomical order and a functional translation that together characterise the local circuit. Thus understanding the functional translation of normal visual circuits will enable a more complete picture of how, when and why such circuits develop.

We undertake a number of in-vivo imaging techniques to probe the functional translations in various local circuits of the mouse visual system. Multi-electrode electrophysiology recordings, two-photon imaging and intrinsic optical imaging (under-development) are employed in conjunction a number of visual paradigms. Classical gratings and sparse noise visual presentations are compared to white and red sub-space noise movies that more closely represent natural scenes.

Our work is revealing functional surrogate markers of the complex underlying anatomy – receptive maps and giving insights as to the nature, form and extent of the functional translations in the early mouse visual processing stream.